Provenancing East Mediterranean cedar wood with the 87 Sr/86 Sr strontium isotope ratio

1 23

Archaeological and AnthropologicalSciences ISSN 1866-9557 Archaeol Anthropol SciDOI 10.1007/s12520-015-0242-7

Provenancing East Mediterranean cedarwood with the 87Sr/86Sr strontium isotoperatio

Sara Rich, Sturt W. Manning, PatrickDegryse, Frank Vanhaecke & Karel VanLerberghe

1 23

Your article is protected by copyright and

all rights are held exclusively by Springer-

Verlag Berlin Heidelberg. This e-offprint is

for personal use only and shall not be self-

archived in electronic repositories. If you wish

to self-archive your article, please use the

accepted manuscript version for posting on

your own website. You may further deposit

the accepted manuscript version in any

repository, provided it is only made publicly

available 12 months after official publication

or later and provided acknowledgement is

given to the original source of publication

and a link is inserted to the published article

on Springer's website. The link must be

accompanied by the following text: "The final

publication is available at link.springer.com”.

ORIGINAL PAPER

Provenancing East Mediterranean cedar wood with the 87Sr/86Srstrontium isotope ratio

Sara Rich1& Sturt W. Manning2 & Patrick Degryse3 &

Frank Vanhaecke4 & Karel Van Lerberghe1

Received: 2 May 2014 /Accepted: 28 April 2015# Springer-Verlag Berlin Heidelberg 2015

Abstract 87Sr/86Sr isotope ratios of cedar wood from forestsin the East Mediterranean have been compiled in order toinvestigate the feasibility of provenancing archaeological ce-dar wood finds.Cedrus sp. forests furnished a great amount ofwood in antiquity, for purposes ranging from ship to templeconstruction, and for fashioning cult statues and sarcophagi.The 87Sr/86Sr signatures of archaeological cedar samples maybe compared with the preliminary dataset presented here tohelp determine the geographic origin of wooden artifacts.Sample sites include two forest areas in the Troodos Massifof Cyprus, five in the Lebanon, and two in Turkey’s TaurusMountains. Sr ratios for wood varieties (i.e., early heartwood,late heartwood, sapwood, and twig wood) demonstrate rela-tive uniformity between the xylem types frequently recovered

from archaeological contexts. As such, this pilot study alsoassesses important issues of archaeological sampling and thegeographical factors that influence Sr uptake in cedar trees ofthis region. While the regional signatures are distinct in mostcases, small sample sizes and range overlap indicate the needfor additional methods to make a case for a certain sourceforest. Alone, this method continues to be best used to dis-prove assumed wood provenances.

Keywords Cedrus sp . Dendroprovenance . EastMediterranean . Strontium isotopes . Timber trade

Introduction

Cedar of Lebanon (Cedrus libani (A. Rich)) timber is oftenregarded as the forest product par excellence of the ancientNear East. Cedar forests and wood alike bore great culturalsignificance, known from at least the Late Neolithic–EarlyBronze Age. Since the Predynastic, Egypt imported cedarwood because it was valued by artisans and technicians forits water, rot, and parasite resistance, and the wood was equal-ly desired by Mesopotamian kings for palace and temple con-struction (Meiggs 1982: 49-87). Although scholars commonlyassume that raw cedar timber was shipped exclusively fromthe forests of Lebanon, there were at least two other areas withcedar variants that were also sought as timber resources: theTaurus and northern Amanus Mountains in Turkey (C. libaniA. Rich, though considered by some as C. libani var.stenocoma (O. Schwarz)) and the Troodos in Cyprus(Cedrus brevifolia (Hook f.) A. Henry) (Kuniholm et al.2007; Lev-Yadun 2008; Hajar et al. 2010; Dagher-Kharratet al. 2007). Likewise, it is highly probable that the forestson the Syrian Coastal Range also would have furnished cedarin antiquity, and there are a few scattered mixed cedar stands

* Sara [email protected]

Sturt W. [email protected]

Patrick [email protected]

Frank [email protected]

Karel Van [email protected]

1 Near Eastern Studies, Katholieke Universiteit Leuven, BlijdeInkomstraat 21, 3000 Leuven, Belgium

2 Malcolm and Carolyn Wiener Laboratory for Aegean and NearEastern Dendrochronology, Cornell University, B-48 Goldwin SmithHall, Ithaca, NY 14853-3201, USA

3 Section Geology, Centre for Archaeological Sciences, KatholiekeUniversiteit Leuven, Celestijnenlaan 200E, 3001 Leuven, Belgium

4 Department of Analytical Chemistry, Ghent University, Krijgslaan281 - S12, 9000 Ghent, Belgium

Archaeol Anthropol SciDOI 10.1007/s12520-015-0242-7

Author's personal copy

in north and central Turkey also thought to be relict (Boydak2003: 232 and refs., Fig. 2; Akkemik 2003: 63) (Fig. 1).

The main aim of this provenance study is to match the87Sr/86Sr ratios of archaeological cedar wood with those ofliving forest trees. In this way, we establish a more completepicture of the use andmobility of cedar wood in the ancient EastMediterranean, while highlighting the role of extra-Lebaneseforests in long-distance trade. As the foundational aspect of thispilot study, the forest wood is examined first to establish region-al 87Sr/86Sr signatures that will help determine ancient woodorigin, and to draw attention to the study’s limitations and somepertinent sampling methods. In particular, we highlight that, BItis only ever possible to disprove a source hypothesis, never toprove one^ from such provenance work (Pollard 2011: 637).

The applications of strontium isotopic analysis abound(Vanhaecke et al. 2009), and it is commonly used as a methodof archaeological provenance (Balcaen et al. 2010 – but, cf.,caveats of Pollard 2011: 634-637). This technique has beensuccessful in provenancing archaeological coniferous woodsin southwest USA, where some of the earliest work on geo-chemical wood provenance was accomplished (Graustein andArmstrong 1983; Gosz and Moore 1989; Durand et al. 1999).English et al.(2001); and Reynolds et al. (2005) provenancedarchaeological timbers from Chaco Canyon, NM, to theirmountain sources by mapping the 87Sr/86Sr ratio in livinggroves, and matching the values thus obtained with results

from the archaeological wood. The method described here isderived from these earlier studies and applied to Cedrus sp.forests in the East Mediterranean.

Strontium is an alkaline earth metal and is present in allrocks, but its concentration varies between petrological types.More specifically, as 87Sr is produced viaβ-decay of 87Rb, the87Sr/86Sr isotope ratio is a function of the rock’s initialrubidium-strontium ratio (87Rb/86Sr) and the time duringwhich these two elements have resided together. For example,older granites with a higher initial amount of 87Rb are charac-terized by a higher 87Sr/86Sr ratio than in many other types ofbedrock. Likewise, newer basaltic volcanic rocks, with lowinitial amounts of Rb, have low 87Sr/86Sr ratios, and sedimen-tary rocks have intermediate ratios (Gosz and Moore 1989;Banner 2003; Slovak and Paytan 2011). Because of the rela-tively high atomic mass of Sr, its isotope ratio remains con-stant in low temperature processes, so plant life growing atopthe bedrock leaches Sr with the same 87Sr/86Sr ratio from thesoil, up-taking it into the stem and providing a measurableisotope ratio, theoretically unique to its underlying lithology(English et al. 2001). However, factors such as local rainfall,groundwater, differential weathering of minerals, and aerosolsalso alter the bioavailable Sr, so plant, animal, and geologicalsignatures are often divergent within the same site (Poszwaet al. 2004; Price et al. 2002). Furthermore, every species ofplant reacts to each of these strontium contributors in a

Fig. 1 Map of the study area withapproximate site locations noted:1, ANT, Antalya-Elmalı; 2, FEK,Feke Düşmüş; 3, SCV, Cyprus A;4, SCV, Cyprus B; 5, TSV,Tannourine-Hadeth; 6, ECF,Horsh-Ehden; 7, BCM, Bcharre;8, THV, Tannourine al-Faouqa; 9,BAR, Barouk

Archaeol Anthropol Sci


different manner, resulting in a strontium isotopic signaturethat is specific to both the species and location (Gosz andMoore 1989; Durand et al. 1999; Degryse et al. 2010).

Materials and methods

Cyprus

In western Cyprus, increment cores of 5-mm diameter weretaken from 24 C. brevifolia trees (two to three cores per tree)on an isolated ridge (site A) in July 2009. Increment cores of5-mm diameter were also taken from 78 trees (two to threecores per tree) in the main Cedar Valley area (site B) inJuly 2007, and 3 more trees were sampled with a larger 10-mm-diameter isotope corer in July 2009 (one core per tree)(Fig. 1). Cores were taken for the dual purposes of the isotopicprovenancing study and dendrochronology, revealing the po-tential for a ≤600-year cedar chronology for Cyprus (Richet al. 2012). Four samples of bedrock were also taken fromeach site directly beneath the area of the sampled trees(Table 1 and Fig. 2).

Of cores extracted from Cypriot cedars in 2007 and 2009,six from each site (A and B) were also selected for the firstround of strontium isotopic analysis via multi-collector induc-tively coupled plasma—mass spectrometry (MC-ICP-MS)

(Balcaen et al. 2010; Vanhaecke et al. 2009). The proceduresfor this type of analysis are reported in Rich et al. (2012), andthe same protocol was employed for all samples discussedhere. Shavings (1–2 g) were removed from the innermost pithring of three samples from each site so that the results wouldindicate the strontium content and 87Sr/86Sr ratio of the earlyheartwood (EH), or the oldest part of each tree (its juvenilegrowth). The same procedure was used on the other half of thesamples (the other three from each site), except that shavingsremoved from these cores represent the late heartwood (LH)to offer a consistent physiological basis for comparison be-tween samples. Additionally, three samples of the mafic dia-base bedrock (1 g) from each site were also submitted to87Sr/86Sr analysis. These 18 samples in total were prepared(acid digestion and chromatographic Sr isolation) and ana-lyzed using MC-ICP-MS at the Department of AnalyticalChemistry at Ghent University. In 2012, three more sampleswere added to this group, two from site A and one from anunknown provenance removed from a cross section of a cedarlog available at the Stavros tis Psokas Forest Station.

Turkey

In 2010, 10 C. libani wood samples from Elmalı, Antalya(Fig. 1), were selected, prepared, and analyzed, using the samemethods as outlined above, to compare with the first Cypriot87Sr/86Sr results. Each sample of the Turkish cedar wood,from cross sections stored at the Malcolm and CarolynWiener Laboratory for Aegean and Near EasternDendrochronology (Cornell University), represented eitherthe sapwood (SW) or heartwood (HW) of each tree, with fivesamples in each category (Table 1). Comparison of the heart-wood and sapwood was undertaken to determine the differ-ence, if any, in 87Sr/86Sr signatures between the two types ofwood for this species, similar to the juvenile versus late heart-wood sub-study of the Cypriot cedars. In 2012, cedar samples(also housed at Cornell) from another Taurus cedar forest,Feke Düşmüş, were added to the study. This site is locatednear the boundary of the Taurus and northern Amanus, northof the city of Adana (Fig. 1). Heartwood from only threerepresentative trees was used, and three of the five samplesanalyzed were taken from the same tree (FEK 12) to demon-strate measurement consistency.

Lebanon

In November 2010, Lebanese cedar and bedrock were collect-ed in the field. Increment cores of 5-mm diameter were takenfrom four forest sites in North Lebanon (Tannourine al-Faouqa, Tannourine-Hadeth (erroneously called Laqlouq inRich et al. 2012), Ehden, and Bcharre), and loose twigs werecollected from Barouk in South Lebanon for comparison(Fig. 1 and Table 1). Bedrock was also collected from each

Fig. 2 The blue box defines the weighted average ±2 * avg. percentsample error, excluding the two outliers of SCV 70A and SCV 6A(both from site B, one LH and one EH), thus defining a Bmain group^of 0.7085–0.7093. However, there is a tendency for data to fall to thelower end of this range, which should be considered when comparingunknown data. That being said, the unknown sample (UL), SCV X, fallswell within this main group, and there is an overall uniformity betweensites A and B and the early and late heartwood samples from each area.The range demonstrates the need to consider these two areas on Cyprus asa single 87Sr/86Sr signature for the purposes of archaeological cedar woodprovenance. AE = site A, EH; AL = site A, LH; BE = site B, EH; BL = siteB, LH; UL = unknown, LH



Table 1 87Sr/86Sr isotope ratios for cedar wood (Cedrus brevifolia and C. libani) and bedrock samples

Wood Sample ID Ring count Type Location, elevation (m) 87Sr/86Sr value (2σ)

Site A, Pafos Forest (Cyprus)

SCV 1108A 598 EH N 35.01778767°, E 32.65582678°, 1230 0.70910 (0.00012)

SCV 1103B 99 EH N 35.01778767°, E 32.65582678°, 1230 0.70832 (0.00016)

SCV 1110B 324 EH N 35.01785078°, E 32.65567582°, 1230 0.70913 (0.00008)

SCV 1113A 374 EH N 35.01811598°, E 32.65561086°, 1230 0.70852 (0.00010)

SCV 1109B 278 LH N 35.0179439°, E 32.65567976°, 1230 0.70881 (0.00025)

SCV 1117A 187 LH N 35.0181537°, E 32.65535471°, 1230 0.70899 (0.00032)

SCV 1119A 164 EH N 35.0181537°, E 32.65535471°, 1230 0.70896 (0.00034)

SCV 1123B 176 LH N 35.01819637°, E 32.65551589°, 1230 0.70886 (0.00020)

Site B, Pafos Forest (Cyprus)

SCV 70Aa 524 EH N 34.99224734°, E 32.68558396°, 1215 0.70808 (0.00022)

SCV 1100A 205 EH N 34.98703689°, E 32.68745237°, 1215 0.70911 (0.00010)

SCV 1102A 338 EH N 34.98668443°, E 32.68744935°, 1215 0.70908 (0.00021)

SCV 6Aa 263 LH N 35.00251617°, E 32.6837495°, 1215 0.70791 (0.00026)

SCV 16A 342 LH N 35.00090517°, E 32.68526209°, 1215 0.70857 (0.00023)

SCV 130B 289 LH N 34.99163806°, E 32.68578814°, 1215 0.70856 (0.00025)

Unprovenanced cross-section of C. brevifolia log from Stavros tis Psokas forest station

SCV X n/a LH n/a 0.70868 (0.00024)

Antalya (Elmalı), Cığlıkara Forest (Turkey)a

ANT 4AH 498 EH N 36.40°, E 29.55°, 1700 0.70835 (0.00008)

ANT 4AS 498 SW N 36.40°, E 29.55°, 1700 0.70810 (0.00009)

ANT 11BH 488 EH N 36.40°, E 29.55°, 1800 0.70848 (0.00014)

ANT 11BS 488 SW N 36.40°, E 29.55°, 1800 0.70894 (0.00006)

ANT 19BH 259 EH N 36.40°, E 29.55°, 1800 0.70849 (0.00022)

ANT 19BS 259 SW N 36.40°, E 29.55°, 1800 0.70839 (0.00017)

ANT 31BH 620 EH N 36.40°, E 29.55°, 1700 0.70857 (0.00010)

ANT 34BH 459 EH N 36.40°, E 29.55°, 1700 0.70854 (0.00010)

ANT 34BS 459 SW N 36.40°, E 29.55°, 1700 0.70848 (0.00012)

ANT 35S 324 SW N 36.40°, E 29.55°, 1700 0.70846 (0.00017)

Feke (DüşmüşOrmanDeposu) (Turkey)a

FEK 6 n/d EH N 37.82°, E 35.91°, 1800 0.71106 (0.00018)

FEK 8 n/d EH N 37.82°, E 35.91°, 1800 0.71539 (0.00014)

FEK 12A n/d EH N 37.82°, E 35.91°, 1800 0.71214 (0.00041)

FEK 12B n/d EH N 37.82°, E 35.91°, 1800 0.71161 (0.00029)

FEK 12B2 n/d EH N 37.82°, E 35.91°, 1800 0.71162 (0.00012)

Barouk (Lebanon)

BAR 1 n/a Twigs N 33.686742°, E 35.698818°, 1787 0.70827 (0.00020)

Bcharre (Lebanon)

BCM 10A 215 EH N 34.242690°, E36.020896°, 1919 0.70833 (0.00066)

BCM 3B 214 EH N34.242671°, E 36.022393°, 1943 0.70827 (0.00246)

Horsh-Ehden (Lebanon)

ECF 2B 150 EH N34.309582°, E 35.997461°, 1635 0.70798 (0.00038)

ECF 3A 147 EH N34.309446°, E 35.997467°, 1644 0.70795 (0.00040)

ECF 5A 138 EH N34.309286°, E 35.997005°, 1646 0.70769 (0.00259)

ECF 1Ab 80 EH N34.309084°, E 35.992761°, 1544 0.70850 (0.00014)

Tannourine-Hadeth (Lebanon)

TSV 2B 31 EH N34.240388°, E 35.922206°, 1507 0.70866 (0.00036)

TSV 7A 71 EH N34.235560°, E 35.921670°, 1590 0.70897 (0.00063)

TSV 8A 109 EH N34.235283°, E 35.921574°, 1593 0.70850 (0.00064)



site except Bcharre, all of which rests on either the massiveJurassic Kesrouane or Mid-Upper Cretaceous Sannine lime-stone formations, with the noted exception of the Tannourine-Hadeth forest growing on Lower Cretaceous Chouf sandstone(Beals 1965). To create dendrochronological time series, firstand second measurements of tree-ring widths from the 45sampled Lebanese cedars were taken with a LINTAB platformand TSAP-Win software at the Wood Science ResearchCentre and Xylarium at the Central Africa Museum inTervuren (Belgium) in 2011 and with a Henson platform andTellervo v. 1.0 software at Cornell’s Tree-Ring Laboratory inFebruary 2012, where the cores are now housed.

Results

Cyprus

Initial results of this experiment were primarily focused on thePafos Forest cedars in Cyprus (Rich et al. 2012). Since thatpublication, three more wood samples have been added to thedatabase for Cyprus (Table 1), reinforcing that paper’s prelim-inary conclusions:

& The Troodos cedar sites have a significantly higher87Sr/86Sr ratio than the bedrock upon which they grow,sharing values much closer to those of the seawater

surrounding the island (Table 1; Rich et al. 2012, Fig. 7).The expanse of the Mediterranean lies only 35 km west ofthe two sites (A=isolated ridge, B=Cedar Valley), so themain westerly wind transport comes from the sea directlyto the Pafos Forest (cf., Evans et al. 2010).

& A sub-study on earlier and later heartwood was also per-formed to determine the variance between respective ringtypes most likely represented in the archaeological mate-rial. With the recent additional data, the second conclusionremains that values are relatively consistent whether thesample is taken from the tree’s early or late heartwood,thereby expanding possibilities for archaeological sam-pling where the process does the least amount of harmvisually and structurally (Table 1 and Fig. 2; see alsoRich et al. 2012, Fig. 6).

& Rich et al. (2012), using dendrochronological cross-datingand the initial strontium isotopic analyses, found that sitesA and B do not differ significantly from each other andshould be treated as a single site.

The recent addition of two more wood samples from site Ademonstrates that the sites’ respective ranges continue to havea significant overlap, with site A’s middle values (mean,0.70884±0.00028 SD; median, 0.70891) being somewhathigher than those of site B (mean, 0.70855±0.00050 SD; me-dian, 0.70856) (Table 2). This similarity becomes even moreapparent when comparing individual samples from both sites

Table 1 (continued)

TSV 9A 85 EH N34.244081°, E 35.927609°, 1482 0.70911 (0.00024)Tannourine al-Faouqa BHead of Valley^ (Lebanon)

THV 3C n/a EH N34.207307°, E 35.927801°, 1727 0.70824 (0.00024)THV 4A 123 EH N34.207061°, E 35.927933°, 1724 0.70848 (0.00015)THV 13A 200 EH N34.207119°, E 35.929650°, 1710 0.70836 (0.00076)

Bedrock Sample ID Type Location, elevation (m) 87Sr/86Sr value (2σ)Site A, Pafos Forest (Cyprus)

SCVA1 Ophiolite N 35.0180°, E 32.6555°, 1230 0.70619 (0.00013)SCVA2 Ophiolite N 35.0180°, E 32.6555°, 1230 0.70558 (0.00008)SCVA3 Ophiolite N 35.0180°, E 32.6555°, 1230 0.70671 (0.00009)

Site B, Pafos Forest (Cyprus)SCV B1 Ophiolite N 34.9870°, E 32.6874°, 1215 0.70525 (0.00008)SCV B2 Ophiolite N 34.9870°, E 32.6874°, 1215 0.70534 (0.00015)SCV B3 Ophiolite N 34.9870°, E 32.6874°, 1215 0.70584 (0.00019)

Barouk (Lebanon)BAR 2 Limestone N 33.686742°, E 35.698818°, 1787 0.70696 (0.00009)

Horsh-Ehden (Lebanon)ECF 1 Limestone N 34.309286°, E 35.997005°, 1646 0.70792 (0.00005)

Tannourine-Hadeth (Lebanon)TSV 1 Sandstone N 34.235560°, E 35.921670°, 1590 0.70736 (0.00013)TSV 2 Sandstone N 34.235560°, E 35.921670°, 1590 0.70739 (0.00012)

Tannourine Al-Faouqa BHead of Valley^ (Lebanon)THV 1 Limestone N 34.207119°, E 35.929650°, 1710 0.70698 (0.00009)

EH early heartwood, LH late heartwood, SW sapwood, n/d no dataa No GPS data available; locations and elevations are approximateb It may be noted that ECF 1A is somewhat anomalous, coming from a lone tree 100 m lower than the rest of the sampled forest, and on a west-facingslope. These or other factors may be driving up its 87 Sr/86 Sr value relative to the other samples from higher altitudes and north-facing slopes



(Fig. 2). The slight tendency of site B’s samples toward thelower end of the Bmain group^ may be the result of subtlegeological variation, or it may be explained by site A’s closerlocation to Chrysochou Bay, visible from its (north)west-fac-ing ridge. Through slightly greater exposure to seawater, siteA may have an increased amount of bioavailable Sr, whichcould have marginally raised that site’s isotopic signature incomparison to that of site B. This being said, the sites are notdifferent enough in range or median values to keep them sep-arate for provenance (Fig. 2; compare with treatment of LSsites: Figs. 5 and 6, and Tables 1 and 2).

Here we also include one unprovenanced cedar sample(SCV X) from a cross section of a felled tree at the Stavrostis Psokas Forest Station (Table 1). This sample acts as aforerunner for testing the possibility of archaeological prove-nance. Not surprisingly, it falls within the rather wide Cypriotrange (0.70791–0.70913) but also well within the main group(0.7085–0.7093), with an 87Sr/86Sr ratio of 0.70868 (Fig. 2and Table 2). To demonstrate an inter-regional application,the sample would also fall within the range for the site onMount Lebanon at Tannourine-Hadeth (0.70850–0.70911;Table 2); however, using external evidence, in this case thefindspot, we can safely conclude that this sample originated inthe Pafos Forest on Cyprus’s Troodos Massif.

Turkey

Two sites were investigated in Turkey: one southwest ofAntalya at Elmalıİşletmesi, Cığlıkara Forest (ANT), and aforest north of Adana at Feke, Düşmüş Orman Deposu(FEK). Rich et al. 2012 found that the signatures of earlyheartwood (EH) and sapwood (SW) are consistent in theANT samples, with the very slight tendency to lower87Sr/86Sr values in the SW, in keeping with observations made

on some other trees species (Watmough and Hutchinson 1996;Prohaska et al. 1998; see Rich et al. 2012, Fig. 8). As with theEH and LH samples of C. brevifolia, the western Taurus dataon C. libani indicate that there is no significant differencebetween 87Sr/86Sr ratios of SW and EH in Cedrus species.

The overall ANT values yield a median and mean of0.70848 (SD=0.00021), which is an intermediate value com-pared to other sites (Table 2). The Turkish cedar signature islikely more reflective of the 87Sr/86Sr ratios of the underlyingbedrock or other local Sr contributors rather than the effects ofthe sea, which lies about 50 km to the south of Elmalı, andsoutherly Sirocco winds may not introduce such a significantamount of seawater or sea-spray (Fig. 1). Furthermore, thetrees are growing on predominantly north- and east-facingslopes, so it is again unlikely that the sea to the south wouldhave a significant impact on the trees’ strontium uptake.

By contrast, not just with Antalyan cedars but all sampledcedar wood, the site of Feke Düşmüş provides the study’shighest 87Sr/86Sr values, with a mean of 0.71236±0.00173(Table 2). This average is so far only supported by five sam-ples taken from three trees, but all five lie well above 0.711,with no range overlap between the other East Mediterraneansites investigated here (Fig. 3 and Tables 1 and 2). This un-usually high value is very promising for archaeological prov-enance. Furthermore, the three samples from the same tree(FEK12; two from the same core, FEK12A) are very closein value and faithfully demonstrate measurement consistency(Table 2). Without geological samples for comparison, we canonly surmise that the sharp increase in 87Sr/86Sr ratios is re-flective of geological changes between Antalya-Elmalı andFeke Düşmüş. This second Turkish site lies near the boundaryof the Eastern Taurus and Amanus instead of in the heart of theWestern Taurus, so it seems likely that the trees there areresponding to the distinct geology of these intersecting

Table 2 Statistical results of thesampled cedar forest sites as theyare separated into groups

Location No. Mean Median Min. Max. Std. dev. Std. error

Cyprus, SCV

Site A 8 0.70884 0.70891 0.70832 0.70913 0.00028 0.00020

Site B 6 0.70855 0.70856 0.70791 0.70911 0.00050 0.00017

Sites A+Ba 15 0.70871 0.70881 0.70791 0.70913 0.00039 0.00010

Sites A+Bb 13 0.70882 0.70886 0.70832 0.70913 0.00027 0.00008

Turkey

ANT 10 0.70848 0.70848 0.70810 0.70894 0.00021 0.00007

FEK 5 0.71236 0.71162 0.71106 0.71539 0.00173 0.00078

Lebanon

LS sitesc 6 0.70833 0.70830 0.70824 0.70848 0.00009 0.00004

ECF 4 0.70803 0.70796 0.70769 0.70850 0.00034 0.00017

TSV 4 0.70881 0.70881 0.70850 0.70911 0.00028 0.00014

a Includes sample SCV XbExcludes outliers SCV 70A and SCV 6Ac LS includes samples from BCM, THVand BAR



mountain ranges. For example, if the FEK trees are growingon or near older metamorphic rocks, or the heterogeneousolistostromes present in the region, then this could increasethe biologically available Sr (Slovak and Paytan 2011: 745).Such a hypothetical scenario could also explain the high stan-dard deviation and error at this site.

Lebanon

The wood samples from Lebanon reveal the lowest 87Sr/86Srratios of the three tested countries/loci (Fig. 3 and Table 2). Ofthe five forest sites examined in Lebanon, three display nosignificant differences in Sr isotopic composition, which isreflective of their similar Mesozoic limestone bedrock.These three sites of Barouk (BAR), Bcharre (BCM), andTannourine al-Faouqa (THV) display a higher 87Sr/86Sr ratioin the wood when compared to the substrate, although thedifference is not as significant as was observed in theCypriot samples (Table 1; Rich et al. 2012, Fig. 7). Other thanaerosols, runoff passing through the calcareous limestone bed-rock could contribute to the bioavailable Sr, thus increasingthe plant to whole-rock 87Sr/86Sr isotope ratio. Due to theirnearly identical Sr isotopic signatures, the sites of Barouk,Tannourine al-Faouqa, and Bcharre cannot be differentiatedfrom one another for the purposes of provenance (medians:THV, 0.70836; BCM, 0.70830; BAR, 0.70827; range,0.70824–0.70848). They are combined in a group character-ized by their common arid limestone environment (LS), withthe lowest SD and SE of the sites presented here (Table 2 andFig. 4). It is also of archaeological interest that the loose twigs

from Barouk display a ratio that corresponds well with theanalyzed LS cores, suggesting, along with the sub-studies ofLH, EH, and SW, that the 87Sr/86Sr values are consistentthroughout a cedar tree’s xylem (Figs. 2 and 4; cf. Rich et al.2012, Figs. 6 and 8).

Of the sites sampled, the bedrock 87Sr/86Sr ratios are sig-nificantly lower than the wood ratios; however, the limestonesubstrate at the forest of Horsh-Ehden corresponds in valuewith that site’s wood samples (Table 1 and Fig. 5). Here wefind the lowest wood 87Sr/86Sr values of all sampled forests,

Fig. 4 The median 87Sr/86Sr values from Barouk (BAR), Bcharre(BCM), and Tannourine al-Faouqa (THV) are perfectly aligned and aretherefore combined into a single arid limestone group (LS) (see Fig. 5).The samples from Tannourine-Hadeth (TSV) demonstrate the highest Srratios in Lebanon, while the Ehden (ECF) samples are identified as thegroup with the lowest strontium isotopic signature

Fig. 3 The Feke cedar is unique to all samples in this study, displaying amedian much higher than any other site tested so far, and clearly wellabove the Antalya samples. This difference in value probably reflects adifferent geological setting of the Feke forest, which is located at theconjunction of the Taurus and Amanus Mountains. Outliers aredetermined using the Tukey two-sided test, with a hinge distance factor(outlier coefficient) of 1.5 and a corresponding o.c. of 3 to identifyextremes

Fig. 5 The grouped cedar and corresponding substrate samples fromLebanon are compared here. Wood from the limestone (LS) group(BCM, BAR, THV) has a much higher Sr signature than its bedrock.The median value of wood from Ehden (ECF) is equal to its limestonesample, the only example of alignment betweenwood and bedrock for theareas discussed in this paper. Wood from Tannourine-Hadeth (TSV) hasthe highest Sr ratios, while its sandstone bedrock has intermediate values



with a median of 0.70796 (mean 0.70803±0.00034 SD)(Table 2). While the other Lebanese forests grow in arid land-scapes, the uniquely humid forest at Ehden (ECF) has a great-er level of soil moisture supporting an equally high level ofbiological diversity (Bachour 2007). It is also the only sam-pled forest with predominantly north-facing slopes, while theothers generally face to the south or west, so this forest is therecipient of cooler, moisture-bearing northerly winds.Furthermore, there is local groundwater at Ehden and seasonalrainwater and runoff at the other sites. Limestone is less likelyto be subject to differential weathering; hence, calcite mineralsfrom the bedrock and the local water supply would likelyproduce a uniform bioavailable Sr composition. This homo-geneity contrasts with the significantly higher wood 87Sr/86Srvalues compared to rock 87Sr/86Sr values at all other sampledsites.

Tannourine-Hadeth is the only cedar forest in Lebanongrowing on sandstone instead of limestone. While the sand-stone samples tested slightly higher than non-ECF limestone(Table 1 and Fig. 5), the wood results are by far thehighest in Lebanon, with its median value (0.70881) be-ing just slightly lower than the non-outlier median forCyprus (0.70886; Table 2 and Fig. 6). Unlike Cypriotcedars, Tannourine-Hadeth’s (TSV) ratios are probablynot driven up by the effects of the sea since the site’selevation is considerably lower than those surrounding it(Table 1), and one of the four trees sampled (TSV 002A)is from an east-facing slope. Instead, TSV cedars may beexperiencing the effects of differential weathering ofnon-quartz minerals, which may have a significant

influence on the bioavailable Sr, possibly increasing theplant to whole-rock 87Sr/86Sr ratios.

Discussion

The groundwork of this investigation has already yieldedsome important implications for the provenance of archaeo-logical cedar wood using strontium isotopic analysis.Additional cedar wood samples were added to the databasesince the publication of the initial results (Rich et al. 2012),and these additions have reinforced earlier findings.C. brevifolia wood largely ignores its ophiolite bedrock andinstead demonstrates a greater influence of the seawater car-ried by prevailing westerly winds, in contrast to the samplesfrom the Taurus and Lebanon ranges, which reflect other geo-logical factors (i.e., substrate, groundwater, weathering, and/or aerosols) (Table 2 and Fig. 5). However, at this time, know-ing some Sr contributors does not seem sufficient to generate aprediction model by which to estimate a forest area’s 87Sr/86Srsignature (Evans et al. 2009). In fact, given the mostly hetero-geneous whole rock to corresponding tree 87Sr/86Sr ratios,wood provenance studies using this method for archaeologicalresearch could focus more explicitly on sampling trees, andless on geological contributors (Laffoon et al. 2012).

Also of direct archaeological importance, a variety of cedarxylem has been examined in this study (early heartwood, lateheartwood, sapwood, and twig wood) from Cyprus, Turkey,and Lebanon, with the conclusion that there is no significantdifference in the 87Sr/86Sr ratio between wood parts of thisspecies (Figs. 2 and 4). The tentative implication is that inves-tigators may be able to analyze wood sampled from any partof the stem, the part of the tree most often used for beams,planks, and other large-scale construction materials, withoutcompromising the accuracy of the isotopic results.Furthermore, because there were often hundreds of years ofgrowth between the pith or early heartwood and the trees’outermost rings, this finding also abates concerns that archae-ological cedar 87Sr/86Sr signatures might differ from rural for-ests today because of modern influencing factors, such aspost-industrial pollutants and other disturbances (Table 1;Guéguen et al. 2012a, b; Drouet et al. 2005; Horn et al.1998; Leonelli et al. 2012). Consistency of stem and twigwood suggests that even branch samples from dunnage, treenails, or firewood could be compared to the regional cedar Srsignatures.

Most archaeological wood from the ancient EastMediterranean and Near East is charred, and it is unknownhow this condition might affect the method’s efficacy (cf.Poole et al. 2002; Lev-Yadun 2008); however, because plantash-based glass has been provenanced using this method(Degryse et al. 2006; Degryse and Schneider 2008; Degryseet al. 2010) it stands to reason that charring would not negatively

Fig. 6 Six regional 87Sr/86Sr signatures for East Mediterranean cedarforests are displayed. The site of Feke in southeast Turkey is excludedbecause its isotope ratios are so high that its range would place thenuances of the other six areas inside its shadow (Fig. 3). Median valuesand interquartile ranges are similar for Cyprus’ Troodos (SCV, excludingoutliers; Table 2 and Fig. 2) and Lebanon’s SS (TSV) sites, but the otherthree areas have distinct medians and interquartile ranges



affect the results for wood either. Many uncharred timbers arealso found in this region preserved in dry or waterlogged envi-ronments. In these conditions, rigorous preparatory measuresshould be taken to separate biostratinomic and diagenetic Sr fromthe sample’s biogenic Sr to ensure that sample signature reflectsits origin and not the circumstances of its use and deposit (Slovakand Paytan 2011; Wilson and Pollard 2002; Pollard and Heron2008: 349-352), although more research is needed to determinethe extent of diagenetic Sr contamination in wood specifically.

Given these caveats, the regional forest 87Sr/86Sr signaturesstatistically demonstrate the possibility of provenancing ar-chaeological cedar woodwith this method in all but two cases.The best odds for specific provenance, given current informa-tion, are with the highest 87Sr/86Sr value at the southeasternTurkish site of Feke Düşmüş (Fig. 3). Should archaeologicaldata fall within this unusually high range, it could beprovenanced to that site with high probability because it isso far removed from the other sampled cedar sites. Likewise,on the other end of the spectrum, should archaeological datafall near the low median value of Horsh-Ehden (ECF; Fig. 6),the artifact could likely be provenanced to that forest area.

Of the other sites, the ratios of C. brevifolia on Cyprus(SCV, Troodos) and C. libani at Tannourine-Hadeth (TSV,Leb SS) in Lebanon represent the next-extreme highvalues (Fig. 6). Sample ranges for the Troodos sites andLebanon’s TSV site are overlapping (although for differ-ent geological reasons), and they have comparable medianvalues. As demonstrated with SCV X (Fig. 2), shouldarchaeological data fall within these overlapping ranges,other provenance methods (historical or archaeologicalcontext, or DNA analysis (Rogers and Kaya 2006), forexample) should be employed to determine the origin.On the other hand, Antalya (ANT, Western Taurus) andthe LS group (BAR-BCM-THV, Lebanon) have distinct,narrow interquartile ranges, which could facilitate archae-ological wood provenance.

One limitation to this study is set by the diminutivesize of cedar forests relative to their ancient expanses(especially in Lebanon and Cyprus), which to some de-gree dictates the number of variables, and accordingly,some sites have small sample numbers. Furthermore, thereare no data yet for the few relict cedars in the SyrianCoastal Range or the Amanus south of Feke. In all cases,the method of provenancing with regional strontium iso-topic signatures provides a probability at best, and ideally,conclusions should be reinforced with other evidence:other geochemical analyses, DNA, dendrochronology, orhistorical and archaeological contexts. However, alreadywe can suggest grounds to exclude some archaeologicalcedar wood provenances, and this preliminary work indi-cates that, with additional larger-scale forest sampling, itmay be possible to further discriminate likely cedar woodsources in the East Mediterranean and Near East.

Acknowledgments This project is supported by the Aegean and NearEastern Dendrochronology Project and the Department of Classics atCornell University, and the Centre for Archaeological Sciences at theKatholieke Universiteit Leuven. Research support was also provided bythe Central Africa Museum’s Laboratory ofWood Biology and Xylariumin Tervuren and the IAP VII Project: Greater Mesopotamia, Research ofits Environment and History. Special thanks to the Belgian-AmericanEducational Foundation and the Flemish Government’s DepartementOnderwijs en Vorming for their generous financial support. We also thankthe Cypriot Department of Forests for permission and assistance (in par-ticular, Andreas Christou)—and especially the forestry officers at theStavros tis Psokas station. Additional thanks are extended to NabilNemer at the Tannourine Cedars Forest Nature Reserve in Lebanon forhis invaluable support, field and otherwise, in 2010. For lab analyses anddata management, the authors thank Kris Latruwe (Ghent) and DennisBraekmans (Leuven). At the Cornell Laboratory in 2009 and 2012, weespecially thank Charlotte Pearson and Peter Brewer.

References

Akkemik Ü (2003) Tree rings of Cedrus libani at the northern boundaryof its natural distribution. IAWA J 24:63–73

Bachour I (2007) Soil profiles and soil analysis of the Ehden and BcharryCedar Forests, final report. American University of Beirut, Beirut

Balcaen L, Moens L, Vanhaecke F (2010) Determination of isotope ratiosof metals (and metalloids) by means of inductively coupled plasma-mass spectrometry for provenancing purposes: a review.Spectrochim Acta B At Spectrosc 65:769–786

Banner JL (2003) Radiogenic isotopes: systematics and applications toearth surface processes and chemical stratigraphy. Earth Sci Rev 65:141–194

Beals EW (1965) The remnant cedar forests of Lebanon. J Ecol 53:679–694

Boydak M (2003) Regeneration of Lebanon cedar (Cedrus libani A.Rich.) on karstic lands in Turkey. For Ecol Manag 178:231–243

Dagher-Kharrat MB, Mariette S, Lefèvre F, Fady B, Grenier-deMarch G,Plomion C, Savouré A (2007) Geographical diversity and geneticrelationships among Cedrus species estimated by AFLP. Tree GenetGenomes 3:275–285

Degryse P, Schneider J (2008) Pliny the Elder and Sr-Nd isotopes: tracingthe provenance of raw materials for Roman glass production. JArchaeol Sci 35:1993–2000

Degryse P, Schneider J, Haack U, Lauwers V, Poblome J, Waelkens M,Muchez Ph (2006) Evidence for glass ‘recycling’ using Pb and Srisotopic ratios and Sr-mixing lines: the case of early ByzantineSagalassos. J Archaeol Sci 33:494–501

Degryse P, Shortland A, De Muynck D, Van Heghe L, Scott R, Neyt B,Vanhaecke F (2010) Considerations on the provenance determina-tion of plant ash glasses using strontium isotopes. J Archaeol Sci 37:3129–3135

Drouet T, Herbauts J, Demaiffe D (2005) Long-term records of strontiumisotopic composition in tree rings suggest changes in forest calciumsources in the early 20th century. Glob Chang Biol 11:1926–1940

Durand SR, Shelley PH, Antweiler RC, Taylor HE (1999) Trees, chem-istry, and prehistory in the American Southwest. J Archaeol Sci 26:185–203

English NB, Betancourt JL, Dean JS, Quade J (2001) Strontium isotopesreveal distant sources of architectural timber in Chaco Canyon, NewMexico. Proc Nat Acad Sci 98:11891–11896

Evans JA,Montgomery J,WildmanG (2009) Isotope domainmapping of87Sr/86Sr biospherevariation on the Isle of Skye, Scotland. J GeolSoc London 166:617–631



Evans JA, Montgomery J, Wildman G, Boulton N (2010) Spatial varia-tions in biosphere 87Sr/86Sr in Britain. J Geol Soc London 167:1–4

Gosz JR, Moore DI (1989) Strontium isotope studies of atmosphericinputs to forested watersheds in New Mexico. Biogeochemistry 8:115–134

GrausteinWC, Armstrong RL (1983) The use of strontium-87/Strontium-86 ratios to measure atmospheric transport into forested watersheds.Science 219:289–292

Guéguen F, Stille P, Geagea ML, Perrone T, Chabaux F (2012a)Atmospheric pollution in an urban environment by tree bark bio-monitoring – Part II: Sr, Nd and Pb isotopic tracing. Chemosphere86:641–647

Guéguen F, Stille P, Dietze V, Gieré R (2012b) Chemical and isotopicproperties and origin of course airborne particles collected by pas-sive samplers in industrial, urban, and rural environments. AtmosEnviron 62:631–645

Hajar L, François L, Khater C, Jomaa I, Déqué M, Cheddadi R (2010)Cedrus libani (A. Rich) distribution in Lebanon: past, present andfuture. C R Biol 333:622–630

Horn P, Hölzl S, Todt W, Matthies D (1998) Isotope abundance ratios ofSr in wine provenance determinations, in a tree-root activity study,and of Pb in a pollution study on tree-rings. Isot Environ Health Stud34:31–42

Kuniholm PI, Griggs CB, Newton MW (2007) Evidence for early timbertrade in the Mediterranean. In: Belke K, Kisslinger EAK,Stassinopoulou MA (eds) ByzantinaMediterranea: Festschrift fürJohannes Koderzum 65. Geburtstag, BöhlauVerlag, Vienna, pp365–385

Laffoon JE, Davies GR, Hoogland MLP, Hofman CL (2012) Spatialvariation of biologically available strontium isotopes (87Sr/86Sr) inan archipelagic setting: a case study from the Caribbean. J ArchaeolSci 39:2371–2384

Leonelli G, Battipaglia G, Siegwolf RTW, Saurer M, di Cella UM,Cherubini P, Pelfini M (2012) Climatic isotope signals in tree ringsmasked by air pollution: a case study conducted along theMont BlancTunnel access road (Western Alps, Italy). Atmos Environ 61:169–179

Lev-Yadun S (2008) Wood remains from archaeological excavations: areview with a Near Eastern perspective. Isr J Earth Sci 56:139–162

Meiggs R (1982) Treesand timber in the ancient Mediterranean world.Oxford University Press, Oxford

Pollard AM (2011) Isotopes and impact: a cautionary tale. Antiquity 85:631–638

Pollard AM, Heron C (2008) Archaeological chemistry. RSC, CambridgePoole I, Braadbaart F, Boon JJ, van Bergen PF (2002) Stable carbon

isotope changes during artificial charring of propagules. OrgGeochem 33:1675–1681

Poszwa A, Ferry B, Dambrine E, Pollier B, Wickman T, Loubet M,Bishop K (2004) Variations of bioavailable Sr concentration and87Sr/86Sr ratio in boreal forest ecosystems. Role of biocycling, min-eral weathering and depth of root uptake. Biogeochemistry 67:1–20

Price TD, Burton JH, Bentley RA (2002) The characterization of biolog-ically available strontium isotope ratios for the study of prehistoricmigration. Archaeometry 44:117–135

Prohaska T, Stadlbauer C, Wimmer R, Stingeder G, Latkoczy C,Hoffmann E, Stephanowitz H (1998) Investigation of elementvariablity in tree rings of young Norway spruce by laser-ablation-ICPMS. Sci Total Environ 219:29–39

Reynolds AC, Betancourt JL, Quade J, Patchett PJ, Dean JS, Stein J(2005) 87Sr/86Sr sourcing of Ponderosa pine used in Anasazi GreatHouse construction at Chaco Canyon, NM. J Archaeol Sci 32:1061–1075

Rich SA,Manning SW, Degryse P, Vanhaecke F, Van Lerberghe K (2012)Tree-ring and strontium isotope signatures of Cedrus brevifolia inCyprus. J Anal At Spectrom 27:796–806

Rogers SO, Kaya Z (2006) DNA from ancient cedar wood from KingMidas’ tomb, Turkey, andAl-Aksa mosque, Israel. Sylvae Genet 55:54–62

Slovak NM, Paytan A (2011) Applications of Sr isotopes in archaeology.In: Baskaran M (ed) Handbook of environmental isotope geochem-istry. Springer, Berlin, pp 743–768

Vanhaecke F, Balcaen L, Malinovsky D (2009) Use of single-collectorand multi-collector ICP-mass spectrometry for isotopic analysis. JAnal At Spectrom 24:863–886

Watmough SA, Hutchinson TC (1996) Analysis of tree rings using in-ductively coupled plasma mass spectrometry to record fluctuationsin a metal pollution episode. J Environ Pollut 93:93–102

Wilson L, Pollard AM (2002) Here today, gone tomorrow. Integratedexperimentation and geochemical modeling in studies of archaeo-logical diagenetic change. Acc Chem Res 35:644–651



Provenancing East Mediterranean cedar wood with the 87 Sr/86 Sr strontium isotope ratio

Documents

Transcript of Provenancing East Mediterranean cedar wood with the 87 Sr/86 Sr strontium isotope ratio